How to Monitor Windows Servers with Prometheus: Metrics, Dashboards, and Alerts

Windows Monitoring Overview

Microsoft Windows is a dominant OS for both desktop and server workloads. Monitoring its core resources—CPU, memory, disk, network, and processes—with Prometheus provides reliable observability.

Metric Reference Model

The model consists of three parts: metric collection, Grafana dashboards, and alert rules, forming a closed‑loop monitoring system.

Metric Collection

CPU : usage (%), DPC queue length, waiting thread queue length, context switches.

Memory : physical & virtual usage (%), page‑file usage (%), paging errors, paged/non‑paged memory.

Disk : space usage (%), free space (%), read/write IOPS, queue length.

Network : inbound/outbound bitrate (bit/s), TCP connection counts (listen, total, non‑established, established), TCP retransmission rate.

Process : per‑process CPU & memory usage (%), handle count, I/O bytes.

Recommended Grafana Dashboards

CPU panel – usage, DPC queue, thread queue, context switches.

Memory panel – physical/virtual usage, page‑file, paging errors.

Disk panel – space usage, free space, IOPS, queue length.

Network panel – traffic rates, TCP connection stats, retransmission.

Process panel – per‑process CPU, memory, handles, I/O.

Top‑N panel – cluster‑wide top 5 nodes for CPU, memory, disk, network.

Default Alert Rules

CPU : usage > 80% for n minutes; waiting thread queue > 2 × CPU cores.

Memory : physical usage > 90% for n minutes.

Disk : space usage > 85% or free space < 15% for n minutes.

Network : established connections > x , non‑established connections > x , TCP retransmission > 5% for n minutes.

Challenges of a Self‑Hosted Prometheus Stack for Windows

Multiple isolated VPCs require separate Prometheus deployments, increasing operational overhead.

Full stack setup (Prometheus, Grafana, Alertmanager) is complex and time‑consuming.

Service discovery for Alibaba Cloud ECS tags is not built‑in; custom Go code is needed to call ECS POP APIs, integrate with Prometheus, and maintain version compatibility.

Open‑source Windows dashboards often lack Windows‑specific optimizations.

No ready‑made alert templates for Windows metrics, requiring manual rule authoring.

Alibaba Cloud Managed Prometheus Solution

Alibaba Cloud Prometheus provides a managed, multi‑cloud service that integrates natively with ECS, supplies out‑of‑the‑box Windows Exporter metrics, optimized dashboards, and pre‑defined alert templates.

Configuration Steps

Install Windows Exporter on each Windows host. Follow the official documentation (https://github.com/prometheus-community/windows_exporter#using-a-configuration-file).

In the Alibaba Cloud Prometheus console, open the ARMS Integration Center and add a Windows component.

Select the ECS environment, bind the Prometheus instance, and provide exporter details:

Exporter name – unique identifier for the Windows host.

Port – the listening port configured for Windows Exporter.

Metrics path – default /metrics.

Scrape interval – collection period in seconds.

ECS tag/value – tags assigned to the ECS instance; Prometheus uses them for service discovery.

Viewing Built‑In Dashboards

Three default dashboards are available:

Windows Detail – shows CPU, memory, disk, and network metrics for a single host.

Windows Process – per‑process CPU, memory, handle, and I/O statistics.

Windows Top‑N – cluster‑wide top 5 nodes for key metrics.

Configuring Alerts

Use the built‑in alert templates or create custom rules based on the metric list above. Detailed steps are described in the Prometheus alert rule documentation (https://help.aliyun.com/document_detail/331981.html).

Custom Metrics

Windows Exporter supports additional collectors such as Active Directory, Container, DNS, etc. To enable them, edit the exporter configuration file (e.g., windows_exporter.yml) to add the desired collectors and restart the windows_exporter service.

Reference Links

Windows Exporter documentation: https://github.com/prometheus-community/windows_exporter#using-a-configuration-file

Alibaba Cloud Prometheus documentation: https://help.aliyun.com/document_detail/122123.html

ARMS Integration Center: https://common-buy.aliyun.com/?commodityCode=arms#/open